§ 01 of 05 — Stage One · Identify
What is this thing, really?
Every digital object in a library carries a fistful of identifiers — an ISBN, an OCLC number, an LCCN, a VIAF cluster, a DOI, a Wikidata Q-number, a local catalog ID. They all claim to point at the same thing. Most of the time, they do. Sometimes they don't. And the work of deciding is older than the web.
A single author, and the cloud of names that points at him
Stephen King has lived in catalogs for sixty years. Twenty-plus national authorities maintain a record for him. Each one assigns its own identifier — its own spelling, its own dates, its own MARC encoding of "this is the same person." VIAF and OCLC's WorldCat Entities API try to keep them in agreement.
- LCNAF · dlcnamesn79018049
- VIAF cluster27066711
- ISNI0000 0001 2103 2683
- FAST · personalfst00041201
- BNF · Francecb11909418n
- DNB · Germany118562525
- BNE · Spain (disagreement)XX1058570
- NDL · Japan00446920
- WikidataQ39829
- OCLC Entity URIE39PBJqv...8mF
- NLA · Australia35867891
- CAOONL · Canada (disagreement)ncf10358912
All of these point to one person — or are supposed to. The dashes are agreement. The red lines are where someone disagrees.
Resolve any identifier — watch the others appear
Paste an identifier from any curator below. The request goes to VIAF, which returns the cluster URI, the preferred heading, and every other identifier VIAF knows about for the same entity. When OCLC's PID Lookup API is provisioned, this same form will additionally return the canonical WorldCat Entity URI and an MD5 content fingerprint.
What this page actually does, and what it doesn't.
- Live VIAF lookups by (curator, id), normalized to a stable JSON shape.
- The full cluster of sameAs identifiers across 20+ national authorities.
- Curator detail panel grounded in a hand-curated registry (lib/curators.ts).
- Bundled fixtures for offline mode; the page works without a network.
- Failure cases — bad input, 404, upstream timeout — rendered as data, not hidden.
- That OCLC's promise of persistence will outlive any one source authority. That promise is institutional, not cryptographic — Stage 5 returns to this.
- That cluster membership is a fact rather than a current best-effort consensus. The clusters split and re-merge as evidence accumulates.
- That the entityMd5 fingerprint is integrity. It is change-detection, not tamper-evidence. We trust OCLC's hash of OCLC's record.
- Two lines on the wall are drawn as 'disagreement' (oxblood, dashed differently) for narrative effect. The actual disagreement set requires Entity Connections API access, which gates on a Meridian subscription.
- The 'OCLC Entity URI' chit on the wall shows a placeholder shape (E39PBJqv…8mF). The real entity URI for King's WorldCat record requires the Entities API. Until access is provisioned, this remains illustrative.
- The entityMd5 field on the resolver card is empty when VIAF is the source. VIAF does not publish a content hash. The annotation says so.
An identifier is a contract.
The word identifier sounds passive — a label, a tag, a barcode — but every identifier in a library is the visible end of a promise. Somebody, somewhere, has agreed to keep resolving it. Without that promise, the string is decoration.
The Library of Congress promises to keep n79018049 pointing at Stephen King. The Bibliothèque nationale de France makes the same promise about cb11909418n. VIAF promises to keep its cluster URI alive, and — here is the deeper promise — to maintain the mapping between all of those national promises. When LC corrects a typo, VIAF notices. When a pseudonym is finally linked to the author behind it, VIAF reflects that. None of this is automatic. It is institutional labor performed for decades, mostly invisibly, by people whose job titles do not say "persistent identifier engineer."
Stage 1 was the easy part: you point at a thing and the system points back. Stage 2 is harder. Which Stephen King? There is more than one, in the wider catalog. There are more than one of nearly everyone. The discipline that resolves them — cluster by cluster, evidence by evidence — is the next room of the museum.
Stage Two · Distinguish — which Stephen King?
You just watched one identifier expand into twenty. Stage 2 starts with the opposite problem: twenty candidates for one person, and the discipline that decides — cluster by cluster, evidence by evidence — which of them is really the same.
Enter the room